update Yu-Tao

Author: Wanyuxiang-code

_data/experience.yml

@@ -1,11 +1,23 @@
 # 学术经历 - 包括研究经历和助教经历，按时间倒序排列
 # 格式：date（时间段）, organization（机构）, role（职位）, type（类型：research/teaching）, description（描述）
 
+- date: "Feb 2026 – Present"
+  organization: "UIUC"
+  role: "Undergraduate Researcher, advised by Prof. Fan Lai"
+  type: "research"
+  description: >
+    Investigating efficient inference for discrete diffusion language models (dLLMs), where inter-step decoding (ISD) pipelines denoising but suffers from a horizon problem due to its flat propose layout.
+    Proposed Tree-Structured ISD with entropy-guided branching to improve tokens-per-forward-pass, and Content-Adaptive Stride that dynamically adjusts propose window width based on local token difficulty.
+    Literature survey and proposal completed; implementation in progress.
+
 - date: "Jun 2024 – Apr 2025"
   organization: "AI-Net Lab, Zhejiang University"
-  role: "Research Intern"
+  role: "Undergraduate Researcher, advised by Prof. Yang Yang"
   type: "research"
-  description: "Working on Graph Foundation Model and Graph-Based In-context Learning. Exploring the intersection of graph neural networks and foundation models."
+  description: >
+    Aimed to improve in-context learning by introducing structural relationships among demonstration candidates and the test query.
+    Proposed ContextG: builds a learnable graph over retrieved examples and the query, uses PageRank-based filtering to select structurally important demonstrations, and encodes them at node/path/subgraph levels to capture multi-hop dependencies.
+    Outperformed SOTA by 2.74% (avg.) on four benchmarks (SST-5, HellaSwag, MRPC, MNLI); best average performance in cross-task transfer. Paper under review.
 
 - date: "Spring 2025"
   organization: "ZJUI"

_data/news.yml

@@ -1,6 +1,9 @@
 # 新闻动态 - 按时间倒序排列，最新的放最上面
 # 格式：date（日期）, text（内容，支持 HTML）
 
+- date: "Feb 2026"
+  text: "🔬 Joined <strong>Prof. Fan Lai's group at UIUC</strong>, working on efficient inference for discrete diffusion language models."
+
 - date: "Oct 2025"
   text: "🎉 Awarded the <strong>National Scholarship</strong> for the academic year 2024-2025!"
 

_data/profile.yml

@@ -27,15 +27,13 @@ about: |
   I'm a junior year undergraduate in Computer Engineering, currently pursuing a dual degree from University of Illinois Urbana-Champaign and Zhejiang University.
 
 about_highlight: |
-  I am exploring the intersection of <strong>machine learning and computer systems</strong>, and I am currently building strong foundations in modern ML workflows and systems fundamentals.
+  My research interest lies in building <strong>autonomous agents</strong> that interact with digital environments. I am particularly excited about:
   <ul>
     <li>
-      Learning the systems challenges behind <strong>LLM training and inference</strong>
-      (e.g., memory, communication, and throughput bottlenecks).
+      <strong>Computer-use agents and GUI grounding</strong> — enabling models to perceive screens, reason over UI elements, and execute actions in real-world software environments.
     </li>
     <li>
-      Exploring <strong>reinforcement learning / multi-agent</strong> workloads from a systems perspective,
-      focusing on scalability and experiment infrastructure.
+      <strong>Efficient inference for language models</strong> — accelerating generation through techniques such as speculative decoding and adaptive computation, with a current focus on discrete diffusion language models.
     </li>
   </ul>
   <strong>Links:</strong><br>